基于二进神经网络的0／1分布系统可靠性分析
系统可靠性的计算依赖于各基本单元的0／1分布关系及其构成的布尔逻辑．本文利用二进神经网络可以完备实现布尔逻辑的特性，提出一种基于二进神经网络的可靠性分析方法．该方法针对每个二进神经元的输入都是0／1逻辑关系的线性组合这一特点，提出并且证明了0／1分布的线性组合的概率分布函数；建立系统功能与布尔函数间的等价关系，将系统转化为相应的二进神经网络；利用线性组合的概率分布函数，通过逐层计算该二进神经网络的0／1输出概率，解决了一般系统的可靠性计算问题．...
Saved in:
| Published in | 自动化学报 Vol. 40; no. 7; pp. 1472 - 1480 |
|---|---|
| Main Author | |
| Format | Journal Article |
| Language | Chinese |
| Published |
广西工学院计算机工程系 柳州 545006
2014
合肥工业大学计算机与信息学院 合肥 230009 合肥工业大学信息与通信工程博士后科研流动站 合肥 230009%合肥工业大学计算机与信息学院 合肥 230009%合肥工业大学计算机与信息学院 合肥 230009 |
| Subjects | |
| Online Access | Get full text |
| ISSN | 0254-4156 1874-1029 |
| DOI | 10.3724/SP.J.1004.2014.01472 |
Cover
| Summary: | 系统可靠性的计算依赖于各基本单元的0／1分布关系及其构成的布尔逻辑．本文利用二进神经网络可以完备实现布尔逻辑的特性，提出一种基于二进神经网络的可靠性分析方法．该方法针对每个二进神经元的输入都是0／1逻辑关系的线性组合这一特点，提出并且证明了0／1分布的线性组合的概率分布函数；建立系统功能与布尔函数间的等价关系，将系统转化为相应的二进神经网络；利用线性组合的概率分布函数，通过逐层计算该二进神经网络的0／1输出概率，解决了一般系统的可靠性计算问题． |
|---|---|
| Bibliography: | Binary neural networks;system reliability;distribution function;linear combination The computing of system reliability relies on the relationship of 0/1 distribution of components and their boolean logic. With the help of the characteristic that binary neural networks can complete the whole boolean function, we propose a method of reliability analysis based on binary neural networks. According to the input of every binary neuron is a 0 or 1 logic variable, we provide and prove the distribution function of the linear combination of 0/1 distribution. Then the equivalent relation between the system function and boolean function is established, and the system is converted to an equivalent binary neural network. As a result, using the distribution function, we can successfully resolve the problem of reliability analysis of general systems by computing the 0/1 output probability layer by layer. 11-2109/TP YANG Juan LU Yang HUANG Zhen-Jin（ 1. School of Computer and Information, Hefei University of Technology, Hefei 230 |
| ISSN: | 0254-4156 1874-1029 |
| DOI: | 10.3724/SP.J.1004.2014.01472 |